Main

CH is a frequent disorder, leading to retardation of mental development and growth if not treated early. In the 1970s, neonatal mass screening programs were started all over the world to guarantee detection and treatment from the 1st wk of life. There is ample evidence that the long-term prognosis has improved since screening was introduced. The growth of early treated patients is essentially normal(1, 2) and mental development has much improved(3–6). However, only a few studies were published in which CH patients were compared with well defined control groups on mental as well as other aspects of neuropsychologic development. To avoid investigator bias and bias caused by the upward secular drift in IQ, cognitive development in CH patients cannot be investigated by comparing the IQ with reference data in the general population. In addition, it appears that in the assessment of neuropsychologic development of CH patients, apart from the evaluation of cognitive function, evaluation of motor skills is extremely important. Many different tools are used for the evaluation of motor development. Therefore, for both aspects of development, evaluation of case-control studies is necessary.

We report an analysis of literature data, to determine: 1) the intellectual development, motor skills, and behavior of early treated CH patients compared with those of controls and 2) which biomedical risk factors for retardation can be defined in these patients.

METHODS

A literature search was performed using MEDLINE data from 1988 up to September 1994 inclusively. Studies which fulfilled the following criteria were included: 1) age of patients and controls ≥5 y and2) control group well defined in the original article. If several studies on the same cohort at different ages were published, the study on the oldest children was included. An exception was made for those studies which described different aspects of development in the different age groups. Comparative studies between CH patients and a reference population not investigated by the authors of the original article, were excluded.

Details of the tests used for the assessment of motor and cognitive functions and of the statistical methods used in the different studies can be found in the original papers.

The following data were collected: number of cases and controls, matching variables, age (distribution), response rate, mean age of treatment, biomedical risk factors (etiology of CH, pretreatment T4, skeletal maturation, T4 dosage); IQ, motor skills, and behavior. Data on full scale, verbal, and performance IQ were pooled, and a weighted mean difference(IQ deficit of patients versus controls) was calculated. If the SD of the IQ distribution was not stated in the original paper, a value of 15 was assumed. Data on the risk factors and on the motor skills and behavior could not be pooled, because of the many different definitions and investigation tools used; therefore, only descriptive comparisons were made.

RESULTS

Studies Included

Seven studies (675 CH patients and 570 controls) fulfilled the inclusion criteria(3, 7–14). The study characteristics are summarized in Table 1. The studies are named after the screening area. The patient group described in the North Thames study is also part of the UK study.

Table 1 Study characteristics
Full size table

Cognitive Functions

Table 2 summarizes the overall data on intellectual functions. In all studies the IQ, the global IQ as well as the verbal and performance IQ, is lower in the CH group than in the control group (deficit of the mean full scale IQ 4-11.8), although the difference does not reach significance in all studies. Meta-analysis of the pooled data showed a significant deficit of the IQ of patients compared with controls (full scale, 6.3; verbal, 6.4; performance, 6.1).

Table 2 Cognitive function in patients and controls
Full size table

Motor Skills

Motor skills are reported in four of the seven studies; none of the groups used the same study tool. Motor functioning of CH patients differed significantly from the control group in all studies (Table 3). Motor impairment seemed to be more pronounced in balance and fine motor function than in gross motor function.

Table 3 Motor skills
Full size table

Behavior

The Norwegian study evaluated behavior by the Richmann behavior checklist. No statistically significant differences were found between CH patients and controls. The Achenbach Child Behavior Checklist was used in the New England study. This study, too, could not detect significant differences.

Biomedical Risk Factors for Disturbance in Cognitive Function(Table 4)

Table 4 Cognitive function (full scale IQ) and risk factors
Full size table

Initial low T4. Four studies found a significant correlation between the IQ and the pretreatment T4. However, the UK study found that this relation was discontinuous with a threshold at 42.8 nmol/L.

Skeletal maturation. The Toronto study compared the intrauterine hypothyroid group (defined as a skeletal development at time of diagnosis ≤36 wk of gestation) with the postnatal hypothyroid group; the intrauterine hypothyroid group had a significantly lower IQ. The Quebec study(at the age of 5 y) shows a significantly lower IQ in those CH patients with an area-of-the-knee epiphysis <0.05 cm2 compared with a control group.

Etiology. Four studies assigned CH patients to subgroups defined by etiology (agenesis, dyshormonogenesis and hypoplasia/ectopia). The Toronto study and the Dutch study compared these subgroups with the control group. Both studies found in the agenesis group a significantly lower IQ. The Quebec study found at the age of 5 y no significant difference between the CH subgroups. The Norwegian study found a significantly lower performance IQ in the agenesis group compared with the ectopic group.

Initial T4 and skeletal maturation. The Quebec study (at the age of 12 y) defined a subgroup with a severe form of CH based on initial T4 and skeletal maturation. Patients with a severe form of CH had a significantly lower IQ than the control group.

Age at start of treatment. In the North Thames study, IQ tended to be lower in the group treated after the age of 30 d in comparison to the group treated from earlier age. Neither the UK study nor the Dutch study found a significant correlation between IQ and age at start of treatment.

T4 dose. In the UK study, no correlation between T4 dose and IQ could be demonstrated. The Norwegian study showed a significant correlation between T4 dose at the age at 1 y and the verbal IQ at the age of 6.

Multivariate analyses. The Toronto study investigated the percentage of variance accounted for by the following variables: age, sex, parental IQ, social status, etiology, bone age, treatment age and age of normalization of T4. The overall p level was <0.05 for all fields of development (language, visuomotor, motor, memory, spatial ability, and auditory discrimination). Parental IQ, bone age, and treatment age accounted for a significant proportion of the variance in language.

The UK study could not detect a relationship between IQ and age at start of treatment, T4 dose, and plasma T4 during the first year of life. The North Thames study (at the age of 5 y) investigated in a multivariate model the independent contribution of initial T4, age at start of treatment and social class to IQ. T4 and social class were independently and significantly associated with IQ.

The Norwegian study assigned CH patient to subgroups defined by mean T4 level in the age group of the 1st y, 1-2 y, 2-4 y, and 4-6 y. The T4 during the 1st y as well as during the age period of 1-2 y was significantly correlated with the 6 y verbal IQ in an analysis of covariance(with sex, social class and initial T4 as covariates). The low T4 group (<129 nmol/L) had a lower verbal IQ. No other significant correlations were found.

Biomedical Risk Factors for Disturbance in Motor Skills

Initial low T4. The Dutch study compared subgroups assigned by initial T4 with the control group; the initial low T4 group performed significantly more poorly. The North Thames study(at the age of 5 y) found a trend toward better motor skills with higher initial T4.

Skeletal maturation. In the Toronto study children in the intrauterine hypothyroid group (definition see above) had significantly poorer motor skills than the postnatal group.

Etiology. The Toronto study showed that the athyroid group rated significantly more poorly, compared with the other etiology groups. The Dutch study compared the three etiology groups with the motor skills of the control group. The agenesis group scored significantly more poorly.

Age at start of treatment. The Dutch study found no significant correlation between motor skills and age at start of treatment. In the North Thames study, the motor skills tended to be lower in the group treated after the age of 30 d.

Multivariate analyses. The Toronto study investigated the percentage of variance accounted for by the following variables: age, sex, parental IQ, social status, etiology, bone age, treatment age and age of normalization of T4. Etiology and bone age accounted for a significant proportion of the variance in motor skills.

The North Thames study investigated in a multivariate model the independent contribution to motor skills of initial T4, age at start of treatment and social class. Only initial T4 was independently and significantly associated with the eventual motor skills.

DISCUSSION

Each year millions of neonates all over the world are screened for congenital hypothyroidism. Early detection and treatment have much improved the prognosis for these children, although individual risk factors may lead to a low IQ and poorer motor skills.

We report an analysis of literature data of seven studies in which CH patients (n = 675) are compared with well defined concurrent control groups (n = 570) investigated by the authors. We excluded other studies for two reasons. 1) Cases as well as controls have been studied by the same investigator, thus avoiding bias. 2) Use of population data as reference may lead to bias as a result of the upward secular shift in the outcome of IQ tests. We have chosen for the most recent published studies, to evaluate a follow up as long as possible. During long term follow up, various aspects of development can improve(11, 15), although noncompliance can influence the results(16).

All studies we analyzed show a trend toward lower IQ and-as far as studied by the authors-poorer motor skills in CH patients compared with controls; meta-analysis showed the deficit to be significant. The most important independent risk factor for the eventual outcome is the severity of CH(defined by initial T4 at the moment of diagnosis and skeletal maturation). Several studies which did not fulfill our inclusion criteria found similar risk factors(17–20). Treatment variables such as age at start of treatment, T4 dose and plasma T4 during treatment do not seem to be an important additional risk factor for the eventual cognitive development in CH found in neonatal screening. The optimal dose of T4 was discussed by various authors(10, 21–23); from the so far published data as well as from the studies analyzed in this article(8, 9) no conclusive information is available. This aspects remains subject for further study.

The results suggest that at least part of the brain damage in patients with CH is caused in utero and cannot be prevented by early treatment. Tillotson et al.(8) estimate that without a screening program 40% of patients with severe hypothyroidism will show impairment of a degree likely to require special education, whereas in early treated CH this percentage is 10.

To assess the amount of prenatal damage to the brain, one would need to have insight in the fetal T4 level. Fetal T4 level is dependent on the production of T4 by the fetus itself as well as on the amount of maternal T4 transferred to the fetus. At approximately 14 d after birth, all maternal T4 transferred has been used by the fetus(21). After that time, the measured T4 will have been produced by the infant itself. Therefore, T4 at diagnosis, which usually takes place after 14 d of age, will be a good indicator of fetal production.

At present there is no good indicator for the amount of maternal T4 that is being transferred to the fetus. It is likely that this amount may vary greatly, because in patients with severe CH it has been shown that cord serum concentrations range from 35 to 70 nmol/L(24). Skeletal maturation is greatly dependent on T4 levels and can therefore be seen as a combined measure of fetal T4 production and maternal T4 transferred to the fetus. The results of Glorieux et al.(3, 14) suggest that a combination of diagnostic T4 and skeletal maturation may be the best predictor, currently available, for later development.

Two study groups designed a predictive model based on individual risk factors. Glorieux et al.(25) predicted in 10 out of 13 patients correctly an IQ < 90 in severe CH, defined as a low initial T4, and retardation of skeletal development (seeTable 4). According to a prediction based on the logistic model of the UK study(8), children with an initial T4 > 47 nmol/L are unlikely to suffer impairment of more the five IQ points. A meta-analysis of data derived from the various studies could probably be the basis for a further refinement of such predictive models for a poor outcome in the individual patient. However, the variability of the definitions used (initial T4, skeletal maturation, treatment parameters) and the diversity of tests for evaluation of the motor development makes such an analysis impossible at this moment.

We conclude that CH, despite early detection and treatment, results in a significant IQ deficit of approximately 6 points. The severity of CH(agenesis, low initial T4, retardation of skeletal malformation), rather than treatment variables, appear to be the most important individual risk factor. Although several biomedical risk factors have been well investigated in many studies, the individual weight is not precisely known. New recommendations of the American Academy of Pediatrics, the American Thyroid Association, and the European Society for Pediatric Endocrinology should help future investigators to design their follow up study in such a way that studies can be compared with each other.